Explore Deep Feature Learning to Power Equipment Monitoring and Defect Detection

Abstract

The research of power equipment defect detection based on image feature has become a hot issue nowadays. In order to solve the problems of low efficiency and accuracy in traditional power equipment defect detection methods, a defect detection method of power metering equipment based on image deep learning is proposed in this work. We train the deep feature learning network model and obtain the optimal solution of network weights in right of training. The association rules are designed and the defect detection mechanism is designed in combination with the collected meter reading dataset. Based on the designed deep network model, defects are identified with the preprocessed images. In the meantime, in order to reduce the power consumption and time delay of data transmission in the process of defect recognition, we introduce the idea of edge computing, so that part of the defect recognition tasks can realize end-to-end intelligence while taking images. Experimental results show that the proposed method can improve defect detection capability and guarantee the normal operation of power metering equipment largely.